Climate intervention on a high-emissions pathway could delay but not prevent West Antarctic Ice Sheet demise

The study addresses whether solar radiation modification (SRM), implemented via stratospheric aerosol injections, can prevent or delay a potential marine ice-sheet instability and collapse of the West Antarctic Ice Sheet (WAIS). SRM is proposed as a means to offset global warming quickly, but its regional impacts, governance, ethical implications, and side effects are uncertain. The urgency arises because the window to limit warming to well below 2 °C is closing, and delayed emissions reductions increase committed warming. WAIS is thought to be vulnerable to tipping via self-sustained grounding-line retreat, yet it remains unclear if current observed retreat reflects an ongoing instability or responses to present climate variability. Given the severe risks from sea-level rise and potential future consideration of SRM, the paper investigates how SRM would influence Antarctic ice-sheet mass balance and sea-level contributions under different greenhouse gas emissions pathways.

Prior research has shown that SRM could, in principle, halt anthropogenic warming rapidly, but the regional consequences, particularly for hydrology, atmospheric chemistry, circulation, and climate variability, remain highly uncertain. Model studies have explored SRM’s global effectiveness and highlighted potential side effects, including altered monsoons, temperature gradients, circulation changes, and unaddressed CO2-driven impacts like ocean acidification. Specific research on polar ice sheets under SRM is limited and has mainly focused on atmospheric and oceanic conditions around Greenland and Antarctica without explicit ice-dynamic modeling. Some unconventional interventions (for example, artificial mass deposition or subsea walls) have been explored. The stability of Antarctic grounding lines under current conditions is debated, with studies suggesting both potential commitment to regional collapse and indications of current stability. Overall, there is a gap in explicit, long-term Antarctic ice-sheet dynamical simulations under SRM scenarios, which this study addresses.

The authors use the Parallel Ice Sheet Model (PISM v1.2) in shallow shelf–shallow ice hybrid mode to simulate Antarctic Ice Sheet (AIS) evolution under multiple climate and SRM scenarios from 1860 to year 3000. PISM is run primarily at 16 km horizontal resolution with selected sensitivity simulations at 8 km and 4 km to assess resolution dependence. The model ensemble varies basal sliding and till friction parameters and employs the PICO basal melt parameterization with multiple heat exchange coefficients to cover uncertainty in ice-shelf melt sensitivity. Calving is represented heuristically via an eigencalving law, and floating ice thinner than 10 m is removed; present-day ice-shelf extent is imposed as a maximum limit. Climate forcing is derived from the HadGEM2-ES Earth system model for RCP 2.6, RCP 4.5, and RCP 8.5, and for SRM scenarios in which SO2 is injected into the stratosphere to maintain global mean near-surface temperature at approximately 1.5 °C above pre-industrial. Original GEO2.6 (SRM26), GEO4.5 (SRM45), and GEO8.5 (SRM85) simulations are extended to 2300 following RCP extensions, and, for ice-sheet simulations, anomalies (relative to pre-industrial) of surface mass balance, air temperature, ocean temperature, and salinity are superimposed on present-day RACMO and World Ocean Atlas fields and regridded to the PISM grid. After 2300, the 2270–2300 forcing is cyclically repeated to 3000. SRM scenario design includes SRM85 starting in 2020 (SRM85-20) and later start dates (2040, 2060, 2080: SRM85-40/-60/-80), created by adding post-2020 SRM85-20 anomalies to the RCP8.5 state at the chosen start year, assuming similar subsequent climate responses. Analogous SRM interventions are considered under RCP4.5 (SRM45-20/-40/-60/-80) with lower SO2 injection rates than under RCP8.5. The ensemble is constrained against observations from 1992–2017 of Antarctic sea-level equivalent ice volume change, mass balance, calving and basal melt fluxes, ice thickness, and grounding-line positions. Model spin-up includes a long fixed-geometry thermal initialization under present-day forcing, a 2,000-year pre-industrial spin with anomalies from HadGEM2-ES added to RACMO and WOA fields, and a historical run to 2005 before branching scenarios.

- Under RCP8.5, Antarctic surface air temperature anomalies over Antarctica peak around 13 °C by 2300; subsurface Southern Ocean (400–700 m near ice fronts) warms by ~1 °C by 2100 and ~3.5 °C by 2300. Widespread surface melt occurs over Ross and Filchner–Ronne ice shelves in the latter half of the 21st century. - In SRM85-20 (start 2020), Antarctic near-surface air temperatures stabilize quickly below ~2 °C, but subsurface ocean continues a committed warming of ~0.5 °C to ~2050 and then increases slowly to 2300. Climatic drivers in SRM85-20 and RCP2.6 are broadly similar, though SRM85-20 exhibits slightly warmer air temperatures and lower ice-shelf surface mass balance. - By 2300 under RCP8.5, WAIS collapse is well underway: WAIS sea-level equivalent (SLE) contribution is ~0.3–0.8 m; AIS SLE is ~0.6–1.1 m. By 3000, WAIS contributes ~2.1–2.8 m and AIS ~2.9–4.0 m. WAIS collapse in RCP8.5 and, in most cases, RCP4.5 is irreversible on millennial timescales. - Model resolution affects timing: at 16 km, Ross Ice Shelf disintegration due to surface melt precedes and contributes to WAIS drawdown, with Thwaites retreat following; at 8–4 km, Thwaites retreats earlier, initiating basin-wide collapse sooner. Long-term SLE by 3000 is similar across resolutions, but pacing differs. - East Antarctica’s George V coast (Wilkes Basin) grounding line remains stable to 3000 across scenarios and resolutions, though Totten and Recovery Basin glaciers lose mass; some East Antarctic mass gains partially offset regional losses. - WAIS stability is mostly ensured only under RCP2.6 (76% of ensemble members) and with early SRM: SRM85-20 (start 2020) or SRM45-20 (start 2040), reflecting similar drivers to RCP2.6. However, even in these cases, committed subsurface warming of ~0.5–0.7 °C by 2300 (regionally up to ~1.0–1.5 °C) can trigger collapse in sensitive ensemble members. - Under current-policy-like pathways (RCP4.5), continuous SRM starting mid-century can sometimes prevent WAIS collapse: SRM45-40 (start 2040) stabilizes grounding lines in 65% of simulations; SRM85-40 triggers long-term collapse in 70% of simulations. - Delaying SRM under RCP8.5 increases collapse probability and sea-level commitment: starting in 2040/2060/2080 yields WAIS collapse probabilities of ~70%/76%/82% and maximum (median) AIS SLE by 3000 of ~1.6 (0.9) m, ~1.9 (1.2) m, and ~2.2 (1.4) m, respectively. SRM starting in 2060 coincides with widespread Ross Ice Shelf surface melt and loss of buttressing. - Median AIS SLE by 3000 across ensembles (Extended Data Table 2): RCP2.6 ~0.31 m (range up to 1.38 m); RCP4.5 ~0.83 m (up to 1.98 m); RCP8.5 ~3.71 m (2.66–3.95 m); SRM85-20 ~0.53 m; SRM45-20 ~0.30 m; SRM85-40 ~0.91 m; SRM45-40 ~0.37 m; SRM85-60 ~0.87 m; SRM45-60 ~0.62 m; SRM85-80 ~1.65 m; SRM45-80 ~0.82 m. - Overall, SRM can delay but not reliably prevent WAIS collapse under high emissions; under intermediate emissions it can reduce risk if deployed early, but with substantial failure risk and side effects.

The simulations indicate that the most robust pathway to prevent a long-term WAIS collapse is rapid decarbonization consistent with RCP2.6, keeping warming well below 2 °C. Early planetary-scale SRM can, in some cases, also avoid triggering marine ice-sheet instability, particularly under intermediate-emissions pathways, but this requires intervention during the first half of the 21st century and still carries substantial risk of failure. Under high emissions (RCP8.5), SRM initiated mid-century largely fails to prevent WAIS collapse; at best it delays the timing, potentially widening the window for adaptation. SRM entails governance, ethical, and physical risks, including regional climate disruptions, does not address CO2-driven ocean acidification, and would likely necessitate centuries to millennia of continued deployment, imposing intergenerational commitments and risks such as termination shock. The study underscores that the emissions pathway’s impact on AIS mass imbalance will not be clearly distinguishable within this century, but delaying mitigation or SRM increases the likelihood of committing to WAIS collapse. The findings support prioritizing rapid emissions reductions while improving understanding of regional SRM impacts before any large-scale deployment is considered.

This work provides the first explicit, millennial-scale Antarctic ice-dynamical assessment under multiple SRM scenarios compared with standard emissions pathways. The key conclusion is that SRM can delay, but under high-emissions trajectories cannot prevent, the long-term demise of the WAIS; under intermediate-emissions pathways, early SRM can reduce or potentially avoid WAIS instability but with significant residual risk. Rapid decarbonization remains the most practical, effective strategy to ensure long-term WAIS stability. Future research should explore coupled ice–ocean–atmosphere modeling frameworks, evaluate polar-targeted SRM designs, assess scenarios that actively reduce temperatures after threshold exceedance, and better quantify SRM regional impacts and risks.

- Stand-alone ice-sheet model: PISM is not interactively coupled to atmosphere–ocean models; missing feedbacks (topographic, freshwater input) could alter timing and magnitude of retreat, likely accelerating loss in coupled frameworks. - Model resolution: Primary simulations at 16 km with selected 8 km and 4 km tests show resolution affects onset timing of instability (earlier Thwaites retreat at higher resolution), suggesting SRM effectiveness may be overestimated at coarse resolution. - Model sensitivity and parameterizations: The chosen set-up exhibits relatively low climate sensitivity compared with some studies; basal melt under ice shelves uses simplified PICO parameterization with uncertain heat exchange coefficients; calving is heuristic; hydrofracturing and cliff failure processes are not included, which could increase vulnerability and expedite retreat. - Initialization and biases: Grounding-line positions and thicknesses deviate from present-day in some critical areas (for example, Thwaites slightly upstream, some East Antarctic regions thicker), potentially biasing retreat timing and regional responses. - Forcing limitations: Post-2300 forcing repeats 2270–2300 conditions, neglecting potential future GHG changes; SRM scenarios assume similar response regardless of start date; HadGEM2-ES has relatively high sensitivity for ocean warming and Antarctic surface mass balance compared with CMIP5 spread. - Scope: The study does not examine SRM strategies aimed at reversing temperatures after threshold exceedance; it focuses on maintaining ~1.5 °C above pre-industrial and continuous deployment assumptions.